The improvement of logistics technology is helpful to reduce the logistical cost,enhance the customer satisfaction, and improve the operation efficiency. The researchof Vehicle Routing Problem is an important measure to improve logistics technology.In recent years, researchers are focused on a mass of branches of vehicle routingproblem, which are derived from real-world applications, and a great deal of researchresults and economical benefits have been gained.The main research of this thesis is Vehicle Routing Problem with soft timewindows, and a model of vehicle routing problem with soft time windows is built bylearning some relevant models. Some usual heuristic algorithms, which are used tosolve vehicle routing problem with soft time windows are introduced. Gene Algorithmand Simulated Annealing algorithm have been learnt, and an Improved GeneAlgorithm based on the two algorithms is built. The Improved Gene Algorithm has agood performance on global search and local search, and some tests show that thisImproved Gene Algorithm is more powerful than sample Gene Algorithm.A multi-depot vehicle routing problem is built on the basis of Solomon’sinstances and Gehring’s benchmark instances, and the problem is solved in this paper.The main step to solve this model is:(1)A new improved method is got by CombiningSweep algorithm and saving algorithm, and this new method is used to transformmulti-depot into single depot problem.(2)Vehicles are distributed by using LINGO,and the lowest allocation decision which met the requirements is obtained.(3)Animproved scanning method is used to allocate demand points, and this improvedscanning method makes use of the capacity of computers maximum, and it protectsinformation of depots not to be destroyed. The results show that this improvedscanning method is more powerful than simple scanning methods.(4) C++and IGAare used to solve the subsystems, and we thus gain the routing and cost of eachvehicle, which shows the superiority of IGA. |